Targeting CD47 and Autophagy Elicited Enhanced Antitumor Effects in Non–Small Cell Lung Cancer

Lung cancer, with non–small cell lung cancer (NSCLC) accounting for 85% of all diagnosed cases, has been the leading cause of cancer-related death worldwide and is increasing in incidence and mortality rate (1, 2). A series of therapeutic interventions, such as platinum-based chemotherapy (3), tyrosine kinase inhibitors (4), and antibodies (5), have been developed to benefit the treatment of NSCLC. However, for the overwhelming majority of NSCLC patients, treatment approaches are scant, and the prognosis still remains poor (6). Therefore, novel and efficient therapeutics for NSCLC are urgently needed.

Clinical trials have revealed that immune checkpoint inhibitors can induce robust antitumor effects and hold promise for treating malignant tumors (7–9). CD47, also known as integrin-associated protein (IAP), is a multifunctional counterreceptor for ligands, such as thrombospondin-1 and signal-regulatory protein-alpha (SIRPα), and affects a range of cellular responses that include cell proliferation, fusion, and migration. Most importantly, CD47 is also known as a key antiphagocytic molecule that renders tumor cells resistant to host immune surveillance (10, 11). Through direct interaction with SIRPα, CD47 acts as a fundamental “don’t eat me” signal and prevents macrophage-mediated phagocytosis in a growing list of malignancy types (12). In this sense, blocking CD47–SIRPα signal transduction by monoclonal antibodies or fusion proteins could increase macrophage phagocytosis of cancer cells. Blockade of the CD47/SIRPα axis by monoclonal antibody is a potential immunotherapeutic strategy for acute myeloid leukemia (13), breast cancer (14), and small cell lung cancer (15). CD47 is also highly expressed on NSCLC cells (16, 17). In the present study, SIRPαD1-Fc, a CD47–SIRPα–blocking fusion protein comprising the first extracellular domain of human SIRPα and the Fc fragment of human IgG1, was engineered and its potential anti-NSCLC efficacy was evaluated.

Autophagy, a key player in microenvironment maintenance, can be induced by many conditions, such as nutrient deprivation, oxidative stress, and drug treatment (18). Although autophagy acts as a “double-edged sword” in tumorigenesis (19), autophagy acts more as a cytoprotective mechanism in tumor therapy (20–22). Many anticancer agents, such as imatinib, cisplatin, vismodegib, and asparaginase, can also induce autophagy in tumor cells, whereas inhibition of autophagy significantly enhances the antitumor efficacies of these agents, indicating that a combination of autophagy inhibitors and antitumor agents could be an efficient therapeutic strategy for malignancy (23–25). Accumulating evidence suggests that blocking CD47 with antibodies confers radioresistance to human normal tissues and cells through inducing a protective autophagy (26–28). Therefore, we hypothesized that combining use of the SIRPαD1–Fc fusion protein to target CD47 with autophagy inhibitors might elicit an enhanced antitumor efficacy.

In the present study, we investigated the therapeutic effects of targeting CD47 with SIRPαD1–Fc fusion protein. Also, we investigated the synergistic antitumor effect of SIRPαD1–Fc in combination with autophagy inhibitors in NSCLC both in vitro and in vivo. Targeting CD47 with SIRPαD1–Fc could elicit potent macrophage-mediated phagocytosis and cytotoxicity against NSCLC cells. During this process, autophagy was dramatically triggered and played a cytoprotective role in NSCLC cells. However, simultaneously targeting CD47 and autophagy significantly increased macrophage-mediated phagocytosis and cytotoxicity against NSCLC cells and showed enhanced antitumor effects in NSCLC xenograft models. Thus, our data demonstrated that simultaneously targeting CD47 and autophagy could elicit enhanced inhibition or even complete elimination of NSCLC in vitro and in vivo, which could be a promising therapeutic strategy for NSCLC.

SIRPαD1–Fc is a recombinant fusion protein based on the first extracellular domain of human SIRPα and the Fc fragment of human IgG1 (Supplementary Fig. S1). To construct the SIRPαD1–Fc expression vector, 57 nucleotides encoding the signal peptide of mouse IgG1 heavy chain were added to the 5′ end of SIRPαD1-coding sequence, a Kozak sequence was added to the 5′ end of the signal peptide sequence, and cloning sites, HindIII and EcoRI, were added to the 5′ and 3′ ends of the resulting sequence, respectively. This designed SIRPαD1 expression cassette sequence was synthesized (Convenience Biotech) and subcloned into the HindIII and EcoRI sites of the pMac-Fc vector (Convenience Biotech, ID: P008-3). The recombinant fusion protein was expressed and purified from Chinese Hamster Ovary (CHO) cells (ATCC, Cat# CCL-61). The Purity of SIRPαD1–Fc fusion protein was above 95%, and the content of endotoxin was below 0.5 U/g.

Reagents were purchased as follows: Rapamycin (Sangon Biotech, A606203), chloroquine and ammonium chloride (NH₄Cl) (Sigma-Aldrich, A9434 and C6628, respectively), N-acetyl-L-cysteine (NAC) and carboxyfluorescein diacetate succinimidyl ester (CFDA SE; Beyotime Institute of Biotechnology, S0077 and C0051, respectively), Cyto-ID autophagy detection kit (Enzo Life Sciences, ENZ-51031-K200), LysoTracker Red dye and MitoSOX Red dye (Invitrogen, L7528 and M36008, respectively). The primary antibodies used for Western blot analyses were purchased as follows: anti-SQSTM1 (Cell Signaling Technology, 8025), anti-LC3 (Cell Signaling Technology, 3868), anti-cytochrome C (Cell Signaling Technology, 11940), anti-PARP (Cell Signaling Technology, 9532), anti–β-actin (Cell Signaling Technology, 3700), anti–caspase-9 (Cell Signaling Technology, 9502), anti-caspase-3 (Cell Signaling Technology, 9665), anti–phospho-Akt (Ser473) (Cell Signaling Technology, 4060), anti-Phospho-mTOR (Ser2448; Cell Signaling Technology, 2971), anti-–Phospho-p70S6 Kinase (Ser371; Cell Signaling Technology, 9208), anti–phospho-4E-BP1/2/3 (Thr45; Abcam, ab68187). The second antibodies used for Western analyses were obtained as follows: horseradish peroxidase (HRP)-conjugated goat anti-rabbit and anti-mouse IgG (MR Biotech, MR-R100, and MR-M100, respectively).

Human NSCLC cell lines A549 and NCI-H1975, and mouse macrophages Ana-1 cell line were obtained in 2015, from Cell Bank of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (Shanghai, China), authenticated under the method of short tandem repeat (STR) fingerprinting by the cell bank. Cells were immediately expanded and stored in liquid nitrogen upon receipt. Each new aliquot was passaged in our laboratory for fewer than six months after resuscitation. All cells were cultured in RPMI-1640 (Corning, 10-040-CVR) containing 10% fetal bovine serum (Capricorn Scientific, FBS-12A), 100 U/mL of penicillin and 100 μg/mL of streptomycin (Beyotime Institute of Biotechnology, C0222) at 37°C in a humidified atmosphere of 5 % CO₂ incubator.

Human ATG7 (siB111124164552) siRNA, ATG5 (siG10726164423) siRNA, and nonsilencing scrambled control (SCR) siRNA (siNO581512211471-10) were purchased from Guangzhou RiboBio Co., Ltd. Human BECN1 (Beclin 1) siRNA and control siRNA (SC-29797) were obtained from Santa Cruz Biotechnology, Inc. A549 cells and NCI-H1975 cells were transfected with siRNA using Lipofectamine 2000 Transfection Reagent (Invitrogen, 11668) following the manufacturer's instructions.

In vitro phagocytosis assay was performed as described previously (16). Briefly, 1 × 10⁵ macrophages were planted per well in glass bottom cell culture dishes (NEST Biotechnology, 801002). According to the manufacturer's protocol, A549 or NCI-H1975 cells were labeled with CFDA SE. Macrophages were incubated in serum-free medium for 2 hours before the adding 2 × 10⁵ of CFDA SE-labeled tumor cells. SIRPαD1-Fc were added and incubated for 2 hours at 37°C. Macrophages were repeatedly washed and subsequently imaged with confocal microscope. The phagocytic index was calculated as the number of phagocytosed CFSE⁺ cells per 100 macrophages.

Cytotoxicity against A549 and NCI-H1975 cells elicited by macrophages was measured by a 6 hours lactate dehydrogenase (LDH) release assay. Pretreated with autophagy inhibitors or siRNA for 12 hours, cells were treated by SIRPαD1-Fc and/or autophagy suppressor or siRNA in the presence of macrophage Ana-1 at the indicated effector:target cell ratio. Then, LDH release was measured by the CytoTox 96 Non-Radio. Cytotoxicity Assay (Promega, G1781) following the manufacturer's instructions.

Human NSCLC A549 and NCI-H1975 cells were incubated with or without SIRPαD1-Fc for 24 hours. After being harvested and processed as described (29), samples were sliced and detected with a JEM 1410 transmission electron microscope (TEM; JEOL, Inc.). The micrographs were taken at 7,000× or 20,000× magnification.

After treatment with SIRPαD1-Fc for 24 hours, A549 and NCI-H1975 cells were rinsed briefly with phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde at room temperature for 10 minutes. Then the cells were permeabilized with 0.2% Triton X-100 for 10 minutes and incubated with rabbit anti-LC3 antibody overnight at 4°C. Subsequently, cells were incubated with FITC-conjugated goat anti-rabbit IgG (Thermo Fisher Scientific Inc., A24532) and 4′, 6-diamidino-2-phenylindole (DAPI) for 1 hour. The images were obtained by confocal microscopy and relative fluoresent intensity was quantified by ImageJ software.

All procedures involving animals were conducted in accordance with the standards approved by Animal Ethical Committee of School of Pharmacy at Fudan University. Six-week-old female BALB/c nude mice (19.5 ± 1.1 g) were subcutaneously injected with 5 × 10⁶ of A549 or NCI-H1975 cells suspended in PBS containing 50% Matrigel Matrix (Coining, 354234) to establish NSCLC xenograft models. Tumor-bearing mice were randomized into 5 cohorts. SIRPαD1-Fc and chloroquine were intraperitoneal injected twice a week and once a day, respectively. The mice treated with cyclophosphamide once a day were served as positive controls. Tumor volume was calculated by the formula: volume = length × width × width/2 (30).

The data were analyzed by GraphPad Prism 5 (GraphPad Software Inc.) and the results were presented as mean ± SD. Comparisons were performed using a two-tailed student t test. P value < 0.05 was considered statistically significant.

The effects of SIRPαD1–Fc on macrophage-mediated phagocytosis of A549 cells and NCI-H1975 cells were detected with confocal microscopy. After treatment with SIRPαD1–Fc for 2 hours, A549 cells and NCI-H1975 cells were efficiently phagocytosed by macrophages (Fig. 1A–D). When compared with isotype control IgG1-Fc, the average phagocytic index in SIRPαD1-Fc–treated cells increased from 5.5 to 29.4 in A549 cells (Fig. 1B) and from 6.6 to 28.6 in NCI-H1975 cells (Fig. 1D). Cytotoxicity against A549 cells and NCI-H1975 cells mediated by macrophages was also determined by LDH release assay. Our findings showed that SIRPαD1–Fc had a negligible direct effect on the viability of A549 cells and NCI-H1975 cells (Supplementary Fig. S2A and S2B), but could significantly induce cell lysis of A549 cells and NCI-H1975 cells in the presence of macrophages (Fig. 1E and F and Supplementary Fig. S2C and S2D).

Taken together, our results indicated that blockade of CD47-SIRPα signaling by SIRPαD1-Fc could elicit potent macrophage-mediated phagocytosis and cytotoxicity against NSCLC cells.

Ultrastructural analysis by TEM was done to observe the formation and accumulation of autophagosomes. After NSCLC cells were treated with SIRPαD1–Fc for 24 hours, an abnormal formation and accumulation of autophagosomes, characterized by double-membrane vesicles, could be observed in the cytoplasm of A549 cells and NCI-H1975 cells (Fig. 2A). Autophagy induction was confirmed by Western blot analysis of the expression of autophagy-related protein LC3 (microtubule-associated protein 1 light chain 3) and SQSTM1 (sequestosome 1). Western blots showed a significant decrease in the expression of SQSTM1 and increase of LC3-II protein in A549 cells and NCI-H1975 cells in a time-dependent manner after treatment with SIRPαD1–Fc (Fig. 2B). Also, Cyto-ID, a dye specific for autophagy, was used to detect autophagic vacuoles, including pre-autophagosomes, autophagosomes, and autophagolysosomes in A549 cells and NCI-H1975 cells after treatment with SIRPαD1–Fc for the indicated times. Similar to the positive control of rapamycin-treated cells, cells exposed to SIRPαD1-Fc increased their punctate fluorescence in a time-dependent manner, with autophagic vacuoles localized in cytoplasm, which confirmed the onset of autophagy (Fig. 2C and D). LC3 immunofluorescent staining was also performed to detect the autophagy induced by SIRPαD1–Fc. Our data (Supplementary Fig. S3A and S3B) showed that treatment with SIRPαD1–Fc significantly (P < 0.01) increased the mean fluorescence intensity of LC3 in A549 cells and NCI-H1975 cells.

To further confirm whether complete autophagy was induced by SIRPαD1–Fc, we examined autophagic flux in SIRPαD1-Fc–treated NSCLC cells. By combined staining of Cyto-ID and LysoTracker, three stages of autophagic flux: formation and accumulation of autophagosomes at 12 hours, fusion of autophagosomes with lysosomes at 24 hours, and degradation of autophagosomes in lysosomes at 36 hours (Supplementary Fig. S3C and S3D) were distinctly observed in SIRPαD1-Fc–treated A549 and NCI-H1975 cells. In addition, chloroquine, a late-stage autophagy inhibitor that could suppress the fusion of autophagosomes with lysosomes and block LC3-II degradation, was applied to further confirm SIRPαD1-Fc–induced autophagic flux. After cells were exposed to SIRPαD1–Fc for various time periods, additional treatment with chloroquine caused a further increase of LC3-II (Fig. 2E and F and Supplementary Fig. S3E and S3F), indicating that SIRPαD1–Fc triggered complete autophagic fluxes that ultimately resulted in degradation of the increased LC3-II in lysosomes.

In summary, these data showed that SIRPαD1–Fc not only induced the initiation of autophagy but also resulted in autophagic flux in A549 cells and NCI-H1975 cells.

Subsequently, we investigated the role of autophagy in macrophage-dependent phagocytosis and cytotoxicity against tumor cells induced by SIRPαD1-Fc. Western blot assay showed that the protein level of LC3-II in the cells cotreated with SIRPαD1–Fc and autophagy inhibitors was higher than that in the cells treated with SIRPαD1–Fc alone (Fig. 3A and B), indicating that both chloroquine and NH₄Cl successfully inhibited SIRPαD1-Fc–induced autophagy in A549 cells and NCI-H1975 cells. Chloroquine or NH₄Cl alone did not significantly increase the phagocytosis and cytotoxicity of macrophages against A549 cells and NCI-H1975 cells (Fig. 3C and Supplementary Fig. S4). Meanwhile, treatment of SIRPαD1–Fc with chloroquine or with NH₄Cl significantly increased SIRPαD1–Fc–induced macrophage-mediated phagocytosis of A549 cells and NCI-H1975 cells (Fig. 3C and Supplementary Fig. S4). LDH release assays showed that SIRPαD1-Fc–induced cytotoxicity against the NSCLC lines A549 and NCI-H1975 mediated by macrophages was greater when autophagy was inhibited (Fig. 3D). Rapamycin, an autophagy inducer, was also used to determine the effect of activation of autophagy in macrophage-mediated cytotoxicity against NSCLC cells (Supplementary Fig. S5A and S5B). Activation of autophagy exhibited negligible effect on the cytotoxicity mediated by macrophages against NSCLC cells.

To further identify the role of autophagy in SIRPαD1-Fc–induced macrophage-mediated phagocytosis and cytotoxicity, we knocked down ATG5 and ATG7, two core autophagy molecules that are necessary for the formation of autophagosomes. Western blot analysis showed that both the siRNA-ATG5 and siRNA-ATG7 selectively reduced the protein expression of ATG5 and ATG7 in A549 cells and NCI-H1975 cells when compared with the nonsilencing scrambled control siRNA-SCR (Fig. 4A and B). After suppression of ATG5 and ATG7 in SIRPαD1–Fc–treated A549 cells and NCI-H1975 cells, macrophage phagocytosis and cytotoxicity against NSCLC cells were significantly enhanced (Fig. 4C and D). We knocked down Beclin 1, a principal regulator in autophagosomes formation, to confirm the role of autophagy in SIRPαD1–Fc–induced macrophage-mediated phagocytosis and cytotoxicity. Western blot analysis showed the selective knockdown of Beclin 1 in siRNA-BECN1–-treated cells (Fig. 4E and F), which significantly increased macrophage-mediated phagocytosis and cytotoxicity (Fig. 4E and F).

Therefore, these data indicated that SIRPαD1-Fc–-induced autophagy played a cytoprotective role in A549 cells and NCI-H1975 cells. Simultaneously targeting CD47 and autophagy could significantly increase macrophage-mediated phagocytosis and cytotoxicity against NSCLC cells.

To investigate the intracellular mechanism of SIRPαD1–Fc–induced autophagy in A549 cells and NCI-H1975 cells, the autophagy-related Akt/mTOR signaling pathway was explored in this study. SIRPαD1–Fc decreased the amount of phosphorylated mTOR in a dose- and time-dependent manner (Fig. 5 and Supplementary Fig. S6). Next, we examined the expression of phosphorylated Akt, an upstream inducer of mTOR, and observed that the phosphorylation of Akt was efficiently inhibited. Furthermore, phosphorylation of p70S6K and 4E-BP1, two downstream substrates of mTOR, was significantly decreased after SIRPαD1-Fc treatment (as shown in Fig. 5A and B,). The autophagy-related Akt/mTOR signaling pathway in xenograft tumors was also examined, and the amount of phosphorylated Akt and mTOR was significantly decreased after SIRPαD1–Fc treatment (Fig. 5C and Supplementary Fig. S7). Thus, SIRPαD1–Fc could downregulate Akt/mTOR signaling and the protein synthesis, which finally activated autophagy in NSCLC cells.

In addition, we examined whether reactive oxygen species (ROS) were involved in SIRPαD1-Fc–induced autophagy in A549 cells and NCI-H1975 cells (Fig. 5D and Supplementary Fig. S8). ROS formation (red fluorescence) was detected approximately 4 hours earlier than the induction of autophagy (green fluorescence) in A549 cells and NCI-H1975 cells after SIRPαD1-Fc exposure. To investigate whether SIRPαD1–Fc–induced autophagy in A549 cells and NCI-H1975 cells was mediated by ROS generation, NAC, a potent antioxidant, was employed to scavenge ROS (Fig. 5E). Pretreatment with NAC (5 mmol/L) was able to counteract SIRPαD1–Fc–induced ROS formation and could partially inhibit SIRPαD1-Fc–induced autophagy. Collectively, these data showed that SIRPαD1-Fc–induced autophagy in NSCLC cells by inactivation of the Akt/mTOR signaling pathway and intracellular ROS formation was involved in the induction of autophagy.

To reveal whether targeting CD47 and autophagy could be a potential therapeutic approach for NSCLC in vivo, A549 xenograft models were established to investigate the synergistic antitumor efficacy of SIRPαD1–Fc and autophagy inhibitor (Fig. 6A). Tumor volume was reduced significantly from day 8 of combined SIRPαD1–Fc and chloroquine therapy (P < 0.001 vs. vehicle). When mice were sacrificed on day 22 and the tumors were resected, the tumor weight in mice treated with SIRPαD1–Fc alone was 357.10 ± 137.81 mg versus 901.43 ± 135.21 mg of the vehicle control (P < 0.0001; Fig. 6A and Supplementary Fig. S9A), and the tumor weight in mice treated with SIRPαD1–Fc in combination with chloroquine was 78.57 ± 49.81 mg (P < 0.001 vs. the cohort treated with SIRPαD1–Fc alone). However, chloroquine did not show antitumor effects in A549 xenograft models when compared with vehicle control. Similarly, in NCI-H1975 xenograft models, tumor weight in mice treated with both SIRPαD1–Fc and chloroquine was 51.43 ± 52.42 mg (P < 0.005 vs. the cohort treated with SIRPαD1–Fc alone), whereas the tumor weight in mice treated with SIRPαD1–Fc alone and the vehicle control were 190.00 ± 76.81 mg (P < 0.001 vs. vehicle control) and 612.00 ± 211.71 mg, respectively (Fig. 6B). One tumor-bearing mouse was tumor free after 24 days of combined treatment of SIRPαD1–Fc and chloroquine (Supplementary Fig. S9B). Taken together, these data indicated that targeting CD47 by SIRPαD1–Fc elicited potent antitumor effects in NSCLC xenograft models and simultaneously targeting CD47 and autophagy could elicit synergistic anti-NSCLC effects.

To investigate the mechanisms of the potent anti-NSCLC efficacy induced by SIRPαD1–Fc and chloroquine, necrosis and apoptosis were thus determined in tumor mass from the tumor-bearing mice. To verify whether blocking the CD47–SIRPα pathway by SIRPαD1–Fc could recruit macrophages to tumor site, CD68, a macrophage specific marker, was used to detect the infiltration of macrophages in transplanted NSCLC (Fig. 6C and D). Histopathologic analysis showed prominent macrophage infiltration of SIRPαD1–Fc–treated tumors. In tumor-bearing mice treated with both SIRPαD1–Fc and chloroquine, macrophage infiltration and tumor cell necrosis increased, indicating that the combination of SIRPαD1–Fc and chloroquine could increase the recruitment of macrophages to the tumor site and this might directly contribute to the inhibition of the developing tumor (Fig. 6 and Supplementary Fig. S10). Inhibiting the CD47–SIRPα pathway also induced autophagy in NSCLC cells in vivo. Ultrastructural analysis and Western blot assays of tumor tissue showed that blocking the CD47–SIRPα pathway by SIRPαD1–Fc induced the accumulation of autophagosomes (Fig. 6E), decreased SQSTM1, and increased LC3-II in transplanted NSCLC cells (Fig. 7A and B). Chloroquine could not only inhibit the autophagy activated by SIRPαD1–Fc, but also significantly increased necrosis in the transplanted tumor tissues (Fig. 7A and B). We then investigated whether ROS and apoptosis were involved in vivo in the killing of tumors after SIRPαD1–Fc and chloroquine treatment (Supplementary Fig. S11). ROS formation (red fluorescence) in NCI-H1975 xenograft tissues was induced after treatment with SIRPαD1-Fc and chloroquine. Western blot analysis of the xenograft tumors showed cleavage of caspase-9, caspase-3 and its substrate PARP, and the release of cytochrome C into the cytosol (Fig. 7 and Supplementary Fig. S12), suggesting that treatment with SIRPαD1–Fc and chloroquine activated the apoptosis pathway.

In brief, these results indicated that simultaneously targeting of both CD47 and autophagy recruited more macrophages to the tumor sites, promoted ROS generation, activated apoptosis-related pathways, and increased necrosis in tumor cells.

To date, immune checkpoint inhibitors that target CD47 fall into three main categories: antibodies to CD47, SIRPα–Fc fusion protein, and SIRPα–antibody fusion protein. Anti-CD47 antibodies, such as B6H12 and Bric126, showed dramatic inhibition of the tumor growth, including ovarian cancer, breast cancer, colon cancer, glioblastoma, and small cell lung cancer, by blocking the transmitted ability of the CD47–SIRPα axis (10, 15, 16). The full extracellular domain of human SIRPα and the Fc fragment of human IgG1-based SIRPα–Fc fusion protein were reported to impair human AML engraftment and dissemination through disruption of SIRPα–CD47 (31). Another study reported that a bispecific fusion protein-based on SIRPα and antibody to CD20 elicited potent elimination of lymphoma cells and caused no significant toxicity in nonhuman primates (11). In this study, we developed a fusion protein SIRPαD1–Fc, consisting of the first extracellular domain of human SIRPα and the Fc fragment of human IgG1, and demonstrated its potent antitumor effects for NSLCL in vitro and in vivo.

A series of immunotherapeutic approaches, including anti–PD-1 therapy and anti–PD-L1 therapy, had previously been established for NSCLC (7, 32). In this study, our results showed that SIRPαD1–Fc therapy could elicit potent anti-NSCLC efficacy in vitro and in vivo. SIRPαD1–Fc could directly target the NSCLC cells and activate macrophage phagocytosis of tumor cells by disrupting the CD47/SIRPα axis “don’t eat me” signal in NSCLC xenograft models. CD47 is also expressed on normal tissues, which could act as antigenic sinks, a potential limitation of treating patients by blocking CD47. To get around this issue, much effort has gone into producing low-affinity/high-affinity bispecific antibodies and SIRPαbodies (11). Meanwhile, new data from the phase I trial of TTI-621 (a SIRPαFc fusion protein) suggest that repeat dosing of TTI-621 overcame the antigen sink while maintaining clinically acceptable platelet levels (33), which could be an effective strategy to circumvent antigenic sinks in other studies. Although many studies have revealed that the therapeutic benefit of blocking CD47–SIRPα interactions in immune competent hosts depends on the myeloid immune subsets, including neutrophils, NK cells, T cells, and dendritic cells (34–36), the primary cell type contributing to effectiveness is macrophages (10, 37–39). Our data here also showed that targeting CD47 by SIRPαD1–Fc elicited potent macrophage-mediated antitumor efficacy in NSCLC. Whether other myeloid immune subsets were activated by this therapy still needs further investigation.

Although increasing evidence shows that blockade of CD47 has potent antitumor efficacy in some solid malignancies, additional approaches to increase efficacy are still urgently needed (40). To further increase the antitumor effects of targeting CD47 by SIRPαD1–Fc in NSCLC, we examined probable mechanisms of resistance to SIRPαD1–Fc treatment. Autophagy, an important mechanism of tumor therapy resistance, is increasingly regarded as a target for synergistic antitumor therapy (41–43). We report here that targeting CD47 could trigger autophagy in NSCLC cells. CD47 deficiency regulates the expression of LC3-II, BECN1, ATG5, and ATG7 to stimulate autophagy and autophagic flux, and conferring radioprotection to normal cells and tissues, which suggests that CD47 blockade could serve as a pharmacologic route to protect normal tissue from radiation injury by modulating autophagy (28). In the present study, our data showed that autophagy was induced through inactivation of the Akt/mTOR signaling pathway, and a complete autophagic flux was also triggered during the treatment, further supporting a crucial role for autophagy in CD47 targeting-based NSCLC therapy.

Up to now, assessments of the role of autophagy in immune checkpoint inhibitors-based tumor therapies were scarce. In this study, we addressed whether autophagy participated in CD47 blockade–based NSCLC therapy. Although autophagy plays a key role in both cell survival and cell death (44), our results strongly support autophagy as a cytoprotective mechanism in this system. We found that inhibition of autophagy while simultaneously providing SIRPαD1–Fc–based NSCLC therapy, by either pharmaceutical inhibitors or genetic approaches, resulted in approximately 90% inhibition or even complete elimination of NSCLC xenograft tumors, indicating that targeting CD47 and autophagy together could be a more effective therapy strategy for NSCLC than targeting CD47 alone.

Besides its regular function in cell microenvironment maintenance, autophagy is associated with the innate and adoptive immune response (45). Autophagy can stimulate the production of immune inhibitory cytokines, such as IL10, TGFβ, and IL27, whereas inhibition of autophagy can be beneficial to the phagocytosis of bacteria by macrophages (46). Our results also indicated that autophagy played a role in resistance to immunotherapy and was involved in immune responses mediated by CD47 blockade in NSCLCs. After suppression of SIRPαD1–Fc–induced autophagy in NSCLC, we observed an increased presence of macrophage infiltration in xenograft tumors, which might lead to apoptosis and growth disadvantage in NSCLC tumors through release of cytochrome C and activation of caspase-9 and caspase-3. ROS generation was also observed in the xenograft tumor cells treated with both SIRPαD1-Fc and chloroquine. A previous study demonstrated that focal regions of macrophage recruitment overlap with regions of enhanced ROS formation (47). Our data indicated that ROS induced by increased macrophage recruitment might partly contribute to NSCLC cell death in the cohort treated with SIRPαD1-Fc and chloroquine. Knockdown of CD47, or binding of CD47 by its ligands, can result in a series of cellular responses, including loss of stem cell characteristics, inhibition of cell proliferation, and spheroid formation, which might promote “eat me” signals that are recognized by macrophages and enhance macrophage phagocytosis (48–50). Here, our data indicated that SIRPαD1–Fc induced signals via CD47 in NSCLC cells and then caused autophagy, a response against cellular stress. Thus, autophagy is probably a mechanism to suppress the “eat me” signals. Our results suggested a probable mechanism of abolishing autophagy to increase the elimination of NSCLC cells by SIRPαD1–Fc in vivo and indicated a novel strategy for NSCLC treatment based on targeting CD47 and autophagy.

Taken together, in the present study, our results demonstrated that targeting CD47 by SIRPαD1–Fc in NSCLC could elicit potent antitumor efficacy. During the treatment, autophagy was triggered via inactivation of the Akt/mTOR signaling pathway and played a cytoprotective role in NSCLC cells. Simultaneously targeting CD47 and autophagy could elicit enhanced macrophage-mediated phagocytosis and cytotoxicity against NSCLC cells and showed enhanced inhibition, or even complete elimination, of NSCLC. A schematic of the mechanisms of the potent anti-NSCLC efficacy induced by simultaneously targeting CD47 and autophagy is presented in Fig. 7C. These data revealed that CD47 was a potential target for use in NSCLC therapy and highlighted the synergistic antitumor effect of simultaneously targeting CD47 and autophagy, providing a scientific basis for further enhancing the antitumor efficacy of immune checkpoint inhibitors.

S. Li is employed at ImmuneOnco Biopharma (Shanghai) Co., Ltd. W. Tian is the founder of ImmuneOnco Biopharma (Shanghai) Co., Ltd. No potential conflicts of interest were disclosed by the other authors.

Conception and design: X. Zhang, S. Wang, W. Tian, D. Ju

Development of methodology: X. Zhang, J. Fan, S. Li, Q. Chen, D. Ju

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Zhang, J. Fan, S. Wang, Y. Li, Y. Wang, J. Luan, P. Song, Q. Chen, D. Ju

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Zhang, J. Fan, S. Wang, Y. Li, Y. Wang, Z. Wang, D. Ju

Writing, review, and/or revision of the manuscript: X. Zhang, J. Fan, S. Wang, Y. Li, D. Ju

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Zhang, J. Luan, P. Song, D. Ju

Study supervision: W. Tian, D. Ju

This work was supported by the National Key Basic Research Program of China under grants 2015CB931800 and 2013CB932502, the National Natural Science Foundation of China under grant 81573332, the Shanghai Science and Technology Funds under grant 14431900200, and Special Research Foundation of State Key Laboratory of Medical Genomics and Collaborative Innovation Center of Systems Biomedicine.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.